Full Length Article

Low cycle fatigue behavior of the extruded AZ80 magnesium alloy underdifferent strain amplitudes and strain rates

Cong Wang, Tianjiao Luo, Yuansheng Yang *Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China

Received 16 March 2016; revised 21 July 2016; accepted 21 July 2016

Available online 8 August 2016

Abstract

Low cycle fatigue behavior of extruded AZ80 magnesium alloy was investigated under uniaxial tension-compression at different strainamplitudes and strain rates. The results show that the extruded AZ80 magnesium alloy exhibits cyclic hardening at strain amplitudes ranging from0.4% to 1.0%, the asymmetry of hysteresis loops becomes increasingly obvious when the strain amplitude increases. Higher strain rates correspondto higher stress amplitudes, high mean stresses and short fatigue life. {10–12} extension twins play a role in the cyclic deformation under higherstrain amplitudes (0.8%, 1.0%). The relationship between total strain energy density and fatigue life can be described by the modified Morrowmodel. The effect of strain rate on the fatigue life can also be predicted by the model.© 2016 Production and hosting by Elsevier B.V. on behalf of Chongqing University. This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Low cycle fatigue; Magnesium alloy; Extruded AZ80; Strain amplitude; Strain rate

Introduction

Due to its low density, high specific strength, and excellentdamping properties, magnesium alloys are widely used in astro-nomical and automotive industries especially when energysaving and environment protection are drawing ever more atten-tion [1,2]. Wrought magnesium alloys exhibit better mechanicalproperties compared with their cast counterparts, so they willfind wide application as structural components. Safe design ofload-bearing components demands information about the cyclicdeformation and fatigue properties. Wrought magnesium alloysusually show tension-compression yield asymmetry related toboth the basal texture that the c-axis is preferentially alignedperpendicular to the extrusion direction or rolling direction[3–7] and the polar nature of twinning that extension twinningcan only be activated by c-axis tension. The yield asymmetrycomplicates the cyclic deformation and fatigue properties ofwrought magnesium alloys.

The cyclic deformation and fatigue properties of wroughtmagnesium alloys have been reported in many studies. Therare-earth element containing extruded magnesium alloys

GW123K [8] and GW103K [9] showed symmetric hysteresisloops under fully reversed strain controlled cyclic deformation,but NZ30K alloy [10] showed asymmetric hysteresis loopsunder similar conditions just as other wrought magnesiumalloys containing no rare-earth element did [7,10,11], the asym-metric hysteresis loops resulted in positive mean stress[7,10–12] which weakened the fatigue resistance, the meanstress first increased and remained constant with increasingcycles in AZ31 magnesium alloy [7,10] but it first decreasedthen increased in AZ61 [11] and AZ31B magnesium alloy [12];cyclic hardening was found in AZ31 [10], AZ31B [3] and AZ61magnesium alloy [11], NZ30K exhibited cyclic softening at lowstrain amplitudes and cyclic hardening at high strain amplitudes[13]; however, GW123K [8] and GW103K [9] showed cyclicstabilization. AZ80 magnesium alloy can serve as structuralmaterial since it has good mechanical properties and relativelylow price [14], but studies on fatigue properties of AZ80 mag-nesium alloy are concentrated on the effect of surface treatment[15,16], fatigue crack propagation [17] or fatigue properties inhigh cycle fatigue regime [18,19], information about the cyclicdeformation and low cycle fatigue properties of extruded AZ80magnesium alloy is lacking [20,21].

Some studies have been carried out to investigate the effectof strain rate on cyclic deformation and fatigue properties ofmagnesium alloys. Duan et al. reported that under strain ratio

* Corresponding author. Institute of Metal Research, Chinese Academy ofSciences, Shenyang 110016, China. Fax: +024 23844528.

E-mail address: ysyang@imr.ac.cn (Y. Yang).

http://dx.doi.org/10.1016/j.jma.2016.07.0022213-9567/© 2016 Production and hosting by Elsevier B.V. on behalf of Chongqing University. This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Available online at www.sciencedirect.com

Journal of Magnesium and Alloys 4 (2016) 181–187www.elsevier.com/journals/journal-of-magnesium-and-alloys/2213-9567

H O S T E D BY

ScienceDirect

of -∞, high strain rates favored {10–12}⟨10–11⟩ twinning–detwinning and inclined to restrict dislocation slip, whichresulted in low hardening rates and long LCF life [22], Begumet al. pointed out that the evolvement of stress amplitudes wasalmost the same under different strain rates, a longer crackpropagation stage resulted in longer fatigue life under highstrain rates [23]. Han et al. [24] investigated fatigue crackpropagation behavior of extruded AZ31 magnesium alloy andfound that high loading frequency corresponded to long fatiguelife. Zeng et al. [25,26] also suggested that higher frequencylead to lower FCP rate for AZ80 and AZ61 magnesium alloys.Since the effect of strain rate on cyclic stress response varieswith loading conditions and it is unclear to what extent thelower FCP rates under higher strain rates would affect the lowcycle fatigue life, this study endeavors to investigate the cyclicdeformation and low cycle fatigue properties of extruded AZ80magnesium alloy under different strain amplitudes and differentstrain rates.

Experimental material and procedures

The material used in this study was extruded AZ80 magne-sium bar with a composition shown in Table 1, its diameter is22 mm and it was extruded at a temperature of 380 °C with theextrusion ratio of 35. Round tensile test specimens with a diam-eter of 5 mm and compressive test specimens with a diameter of8 mm along with dog bone-shaped fatigue specimens withcross section area of 4 × 4mm2 within the gauge section weremachined from the rod using EDM. The axis of these specimenswas parallel to the extrusion direction. The fatigue tests wereconducted on Instron electropuls E10000 servo hydraulictesting machine at room temperature. Low cycle fatigue tests atdifferent strain rates and strain amplitudes were performedunder fully reversed push–pull mode. The strain amplitudeswere 0.4%, 0.6%, 0.8% and 1.0% under strain rate of1 × 10−2 s−1, and the strain rates were 1 × 10−3 s−1, 5 × 10−3 s−1,1 × 10−2 s−1 and 1 × 10−1 s−1 under strain amplitude of 0.8%. Thestrain was measured by an extensometer attached to each speci-men, its gauge length is 10 mm, the knife-edge of the exten-

someter was covered with double-sided tape to protect thespecimen and avoid the relative movement of the specimen andthe extensometer. Before fatigue test, the fatigue specimenswere ground using SiC paper of 5000 grit. Two specimens weretested at each level. The test was stopped when the specimensfractured into two parts, and the corresponding number ofcycles was defined as fatigue life.

The samples for microstructure examination were cut fromthe rod and ground using SiC paper from 800 grit up to 2000grit followed by polishing using 2.5 µm diamond paste. Theetchant of acetic picric solution (4.2 g picric acid, 10 ml aceticacid, 10 ml H2O and 70 ml ethanol) was used to reveal themicrostructure, macrotexture was measured using X-ray dif-fraction. Electron–Back scatter diffraction (EBSD) was used toexamine the microstructure away from the fracture surfaces.

Results and discussion

Microstructure and monotonic mechanical properties

Fig. 1a,b shows the optical microstructure of extruded AZ80along the extrusion direction and perpendicular to the extrusiondirection, respectively, equiaxed grains can be observed withaverage grain size around 8 µm. No twin was found.

The monotonic mechanical properties are tabulated inTable 2. The yield strength and ultimate tensile strength isslightly higher than that reported by Shiozawa et al. [18], whichmay be the result of finer grain size in this study.

The (0002) and (10-10) pole figures are shown in Fig. 2. Theiso-intensity contours are labeled as multiplies of the randomdistribution (mrd) with the maximum intensity around 2 mrd.According to the basal (0002) and prismatic (10-10) polefigures, the texture can be considered as a fiber texture withbasal plane roughly parallel to the extrusion direction.

Effect of strain amplitude

Fig. 3 demonstrates the orientation maps of the fatiguedspecimens on the ED-TD plane at strain amplitudes of 0.4%,

Table 1Chemical composition of AZ80 magnesium alloy (wt%).

Al Zn Mn Mg

7.8–9.2 0.2–0.8 0.12–0.5 Bal.

Fig. 1. Microstructure of extruded AZ80: (a) along extrusion direction, (b) perpendicular to extrusion direction.

Table 2Monotonic mechanical properties of the extruded AZ80 alloy.

Tensile yieldstrength (MPa)

Ultimate tensilestrength (MPa)

Elongation(%)

Ultimate compressivestrength (MPa)

236 343 16 341

182 C. Wang et al. / Journal of Magnesium and Alloys 4 (2016) 181–187

0.8% and 1.0%. At strain amplitude of 0.4%, there is noresidual twin indicating dislocation slip is the main deformationmode. A lot of residual twins can be observed at strain ampli-tudes of 0.8% and 1.0% implying twinning-detwinning plays arole in the cyclic deformation process. Some twins spreadacross the grain and others terminate at the interior of the grain.Most of the residual twins observed are extension twins with amisorientation angle around 86° relative to the original grain asindicated by the point to point misorientation distribution alongline AB in Fig. 7d, inspection of the misorentation angle atother twin boundaries also shows the same result.

Fig. 4 shows the cyclic stress response under different strainamplitudes. The stress amplitude increases with increasing

strain amplitude. Cyclic hardening can be observed under allstrain amplitudes, similar results were observed in extrudedZK60 [27], AZ61 [11], and AZ31 [28,29] alloy. The cyclichardening rates under strain amplitudes of 0.8% and 1.0% aresignificantly higher than that under strain amplitudes of 0.4%and 0.6%. The mechanism accountable for this phenomenon isdiscussed as follows: under small strain amplitudes where dis-location glide is the main plastic deformation mode, the cyclichardening of the material is mainly the result of dislocationmultiplication and their interaction with the second phaseMg17Al12 [30] and the grain boundary; under higher strainamplitudes, apart from the cyclic hardening from dislocation,twinning can be activated during the compressive loading [31],

Fig. 2. (0002) and (10-10) pole figures of extruded AZ80 magnesium alloy.

Fig. 3. Inverse pole figure maps of fatigued specimens under strain amplitudes of (a) 0.4%, (b) 0.8%, (c) 1.0%, (d) point to point misorientation along line ef in.

183C. Wang et al. / Journal of Magnesium and Alloys 4 (2016) 181–187

which can be evidenced by the residual twins observed athigher strain amplitudes, twinning deformation can transformglissile dislocations into sessile ones to impede dislocation slip[2] and residual twins – formed during the compressive loadingand did not disappear during unloading and tensile reloading –accumulate with increasing loading cycles and act as barriers todislocation slip [31,32], moreover, the reduction of twinning–detwinning also makes dislocation slip accommodate moredeformation. All these contribute to the higher cyclic hardeningrates under higher strain amplitudes.

Fig. 4b shows the evolvement of mean stress with loadingcycles, positive mean stress is maintained throughout thefatigue life under all strain amplitudes, but the evolvement ofthe mean stress depends on strain amplitude, when the strainamplitudes are low (0.4%, 0.6%), the mean stresses decreasecontinuously; when the strain amplitudes are high (0.8%,1.0%), mean stresses decrease first then increase, similar trendhas been reported in studies of extruded AZ31 bar [33], and

extruded AZ61 alloy [11]. The evolvement of mean stress isdifferent under high strain amplitudes and low strain ampli-tudes, because the tensile peak stresses increase faster aftersome initial cycles under high strain amplitudes, two reasonsmight be accountable, firstly, after some initial cycles tensilepeak stresses may become high enough to activate the prismaticor pyramidal slip and their interaction with basal dislocationsresults in strong strain hardening [34], secondly, the accumu-lating residual twins impede dislocation slip, which increasesthe stress during tensile loading.

Fig. 5 shows the first two cycles along with the half life cycleof the hysteresis loops under different strain amplitudes. Thehysteresis loops are symmetric under strain amplitude of 0.4%and become asymmetric under strain amplitudes higher than0.6%; the higher the strain amplitude, the more asymmetric thehysteresis loops.

At strain amplitude of 1.0%, the specimen yields as thestress reaches 228 MPa during tensile loading, which is similar

Fig. 4. Cyclic stress response at strain amplitudes ranging from 0.4% to 1% for extruded AZ80 alloy: (a) cyclic stress amplitude, (b) mean stress.

Fig. 5. Hysteresis loops at first, second and half-life cycle under strain amplitudes of (a) 0.4%, (b) 0.6%, (c) 0.8%, (d) 1.0%.

184 C. Wang et al. / Journal of Magnesium and Alloys 4 (2016) 181–187

to the unidirectional tensile yield strength, during compressiveloading, the material shows a strain hardening plateau, which isalso observed at strain amplitudes of 0.8% and 0.6%, the lengthof the plateau increases with increasing strain amplitude. Thestrain hardening plateau was also reported in studies ofextruded ZK60 magnesium alloy [35,36]. The hardeningplateau is the result of extension twinning, direct evidence wasgiven by O. Muránsky et al. using in situ neutron diffraction[36]. During unloading, the linear relationship between thestress and strain maintains temporarily and the stress proceedsnonlinearly with decreasing strain, this is attributed todetwinning process, which is driven by the stress generated bystress redistribution during twinning in the compressive phase[36,37], detwinning generates elastic strain and makes thestress–strain curve nonlinear. So the asymmetry of hysteresisloops is the result of the twinning–detwinning deformationmode activated under high strain amplitudes. In the secondcycle, the work hardening plateau disappears, since residualtwins formed after the first cycle [35], later deformation duringcompressive loading involves the migration of twin boundarieswhich requires smaller force compared with twin nucleation inthe first cycle, so the stress increases gradually. The asymmetryof the hysteresis loop remains in the mid-life cycle indicatingtwining–detwining still exists.

Effect of strain rate

Fig. 6 shows the evolvement of stress amplitude and meanstress at different strain rates under strain amplitude of 0.8%.The stress amplitudes are similar under different strain rates atthe beginning of the cyclic deformation and the specimens

exhibit cyclic hardening under different strain rates. However,cyclic hardening rate increases with creasing strain rate, whichresults in higher stress amplitudes under higher strain rates.Although the evolvement of mean stress is similar at differentstrain rates, mean stress also increases with increasing strainrate. Duan et al. [22] reported opposite result in their study ofAZ31B under strain ratio of -∞: high strain rates favor {10–12}⟨10–11⟩ twinning– detwinning that restricts dislocation slipand result in low stress amplitudes. The strain ratio is differentfrom that used in their study, the length of the strain hardeningplateau of the hysteresis loops which corresponds to twinningdeformation under different strain rates is identical, and theresidual twins away from the fracture surface are also similar asshown in Fig. 7, so it is speculated that the effect of strain rateon the twinning deformation is not remarkable. As twinningand dislocation slip are the plastic deformation modes, it is theinfluence of the strain rate on dislocation slip that leads tovariation of stress amplitude and mean stress with strain rate.

Dislocations slip under the action of external stress andbarriers, some of the barriers can be overcome by thermalactivation, so the shear stress is sensitive to the change in plasticstrain rate, and the stress increases with increasing strain rate.More dislocations will be generated under higher stress, and theincreased dislocation density could impede dislocation slip,which will accelerate the hardening process [38]. Likewise,under cyclic deformation, more dislocations can be generatedunder higher strain rates, which increases the cyclic hardeningrate of the specimens tested at higher strain rates. Since dislo-cation slip and twinning are the main deformation modesduring tension and compression, respectively, tensile peak

Fig. 6. Cyclic stress response at strain rates ranging from 1 × 10−3 s−1 to 1 × 10−2 s−1 for extruded AZ80 magnesium alloy: (a) stress amplitude, (b) mean stress.

Fig. 7. Microstructure away from the fracture surface at strain rates of (a) 1 × 10−2 s−1, (b) 1 × 10−3 s−1 under strain amplitude of 0.8%.

185C. Wang et al. / Journal of Magnesium and Alloys 4 (2016) 181–187

stress increases more than compressive peak stress under higherstrain rates, so mean stress also increases with increasing strainrate.

Fatigue life

The fatigue life of all specimens tested is listed in Table 3.The fatigue life decreases with increasing strain amplitude,strain–life curve of magnesium alloy have a kink point demar-cating different deformation modes related to twinning-detwinning and non-zero mean stress [39], the modifiedMorrow model [7] takes into account the mean stress effect andcan correlate the fatigue life with total strain energy density.The modified Morrow model can be expressed as:

ΔW Nt f* m C= (2)

where Nf is the fatigue life, ΔWt is the total strain energy densityat half-life, m and C are the material constants representing thefatigue exponent and material energy absorption capacity,respectively.

The strain amplitudes used in this study involve twinning–detwinning as deformation mode and result in positive meanstresses. So the modified Morrow model was used in this study.Fig. 8 shows the relation between the fatigue life and the totalstrain energy density. For the material used in this study, fatigue

life can be correlated to the total strain energy densityby this equation. The material constants are m = 0.300 andC = 28.26MJ/m3 as obtained by best fitting the experimentaldata.

The fatigue life increases with decreasing strain rate at strainamplitude of 0.8%. High strain rates produce high stress ampli-tudes and high mean stresses as can be seen from Fig. 6, sincepositive mean stress is detrimental to fatigue resistance and thetotal strain energy density is increased because of the highstrain amplitudes and the high positive mean stresses, so thefatigue life is shorter under higher strain rates. The plastic strainamplitude as a function of the cycle number is plotted in Fig. 9,it first decreases then suddenly increases, which is the result ofthe competition between the cyclic hardening and the fatiguedamage involving crack initiation and crack propagation [12].The number of cycles corresponding to the valley point of theplastic strain amplitude increases with decreasing strain rate.This is in agreement with the result obtained by Begum et al.[23]. The longer fatigue life under lower strain rates is mainlycaused by this prolonged stage of decreasing plastic strainamplitude.

Since the strain rate affects the stress amplitude and meanstress, the modified Morrow model with the obtained materialparameters was used to predict the fatigue life under differentstrain rates. The predicted and experimental data are similar asshown in Fig. 10 suggesting that the model can be used to

Table 3Fatigue life of extruded AZ80 magnesium alloy.

Δε/2 (%) �ε (s−1) Nf

0.4 1 × 10−2 76460.4 1 × 10−2 80230.6 1 × 10−2 17450.6 1 × 10−2 20200.8 1 × 10−2 4240.8 1 × 10−2 3971.0 1 × 10−2 1691.0 1 × 10−2 1450.8 1 × 10−3 6500.8 1 × 10−3 6380.8 5 × 10−3 4360.8 5 × 10−3 4030.8 1 × 10−1 3900.8 1 × 10−1 253

Fig. 8. Relationship between total strain energy density and fatigue life.

Fig. 9. Plastic strain amplitude under different strain rates at strain amplitudeof 0.8%.

Fig. 10. Experimental result of fatigue life and that predicted by the modifiedMorrow model under different strain rates.

186 C. Wang et al. / Journal of Magnesium and Alloys 4 (2016) 181–187

predict the effect of strain rate on the fatigue life of this alloy,although further investigation is needed.

Conclusions

Low cycle fatigue behavior of extruded AZ80 magnesiumalloy under different strain rates and strain amplitudes wasexamined in this study, several conclusions are drawn:

1 The extruded AZ80 magnesium alloy exhibits cyclic hard-ening at strain amplitudes ranging from 0.4% to 1.0%. Withthe increase of strain amplitude, the asymmetry of hysteresisloops becomes increasingly obvious. {10–12} extensiontwins play a role under higher strain amplitudes (0.8%,1.0%).

2 High strain rates produce high stress amplitude, high meanstresses and low fatigue life.

3 The relationship between the fatigue life and total strainenergy density under different strain amplitudes can bedescribed by the modified Morrow model, and the fatiguelife predicted by the model under different strain rates isconsistent with the experimental result.

Acknowledgments

This work was financially supported by the National BasicResearch Program of China (No.2013CB632205).

References

[1] Z. Yang, J.P. Li, J.X. Zhang, G.W. Lorimer, J. Robson, Acta Metall.Sin.(Engl. Lett.) 21 (2008) 313–315.

[2] B. Colleen, M. Barnett, Advances in Wrought Magnesium Alloys:Fundamentals of Processing, Properties and Applications, WoodheadPublishing, Philadelphia, 2012.

[3] C.J. Geng, B. Wu, X.X.H. Du, Y.D. Wang, Y.D. Zhang, F. Wagner, et al.,Mater. Sci. Eng. A 560 (2013) 618–626.

[4] S. Dong, Y.Y. Jiang, J. Dong, F.H. Wang, W.J. Ding, Mater. Sci. Eng. A615 (2014) 262–272.

[5] S.L. Zheng, Q. Yu, Y.Y. Jiang, Int. J. Fatigue 47 (2013) 174–183.[6] C.F. Gu, L.S. Toth, M. Hoffman, Acta Mater. 62 (2014) 212–224.[7] S.H. Park, S.G. Hong, B.H. Lee, W. Bang, C.S. Lee, Int. J. Fatigue 11

(2010) 1835–1842.[8] S.M. Yin, S.X. Li, J Mater. Sci. Technol. 8 (2013) 775–780.[9] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Mater. Sci. Eng. A 575 (2013)

65–73.

[10] S.H. Park, S.G. Hong, W. Bang, C.S. Lee, Mater. Sci. Eng. A 3 (2010)417–423.

[11] J.B. Jordon, J.B. Gibson, M.F. Horstemeyer, H. El Kadiri, J.C. Baird, A.A.Luo, Mater. Sci. Eng. A 23 (2011) 6860–6871.

[12] X.Z. Lin, D.L. Chen, Mater. Sci. Eng. A 1–2 (2008) 106–113.[13] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Mater. Design 64 (2014)

63–73.[14] Z.T. Wang, Y.Q. Yang, B.C. Li, Y.H. Zhang, Z.H. Zhang, Mater. Sci. Eng.

A 582 (2013) 36–40.[15] P. Zhang, J. Lindemann, Scripta Mater. 6 (2005) 80–86.[16] P. Zhang, J. Lindemann, C. Leyens, J. Mater. Process. Tech. 3 (2010)

445–450.[17] R.C. Zeng, Y.B. Xu, W. Ke, E.H. Ha, Mater. Sci. Eng. A 2 (2009) 80.[18] K. Shiozawa, T. Kashiwagi, T. Murai, T. Takahashi, Procedia Eng. 1

(2010) 183–191.[19] Y. Uematsu, K. Tokaji, M. Matsumoto, Mater. Sci. Eng. A 517 (2009)

138–145.[20] H. Zenner, F. Renner, Int. J. Fatigue 12 (2002) 1255–1260.[21] Y. Li, J. Jiang, J. Yang, J. Aeronaut. Mater. 33 (2014) 33–39.[22] G.S. Duan, B.L. Wu, X.H. Du, X. Zhao, Y.D. Zhang, L. Zuo, et al., Mater.

Sci. Eng. A 11 (2014) 11–22.[23] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Int. J. Fatigue 31 (2009) 334–343.[24] Z.Y. Han, H.Y. Luo, C.K. Sun, J.R. Li, M.K. Papaelias, C. Davis, Mater.

Sci. Eng. A 597 (2014) 270–278.[25] R.C. Zeng, E.H. Han, W. Ke, Int. J. Fatigue 36 (2012) 40–46.[26] R.C. Zeng, W. Ke, E.H. Han, Int. J. Fatigue 31 (2009) 463–467.[27] Q. Yu, J. Zhang, Y. Jiang, Q.Z. Li, Int. J. Fatigue 36 (2012) 47–58.[28] S. Kwon, K. Song, K.S. Shin, S.I. Kwun, T. Nonferr. Metal. Soc. 20

(2010) 533–539.[29] C. Geng, B. Wu, X. Du, X.H. Du, Y.D. Wang, Y.D. Zhang, et al., T.

Nonferr. Metal. Soc. 23 (2013) 1589–1594.[30] W. Tang, E.H. Han, Y.B. Xu, L. Liu, Acta Metall. Sin. 11 (2005) 91–98.[31] W. Wu, P.K. Liaw, K. An, Acta Mater. 85 (2015) 343–353.[32] Y.J. Wu, R. Zhu, J.T. Wang, W.Q. Ji, Scripta Mater. 63 (2010) 1077–

1080.[33] S. Hasegawa, Y. Tsuchida, H. Yano, M. Matsui, Int. J. Fatigue 29 (2007)

1839–1845.[34] N. Bertin, C.N. Tomé, I.J. Beyerlein, M.R. Barnett, L. Capolungo, Int. J.

Plasticity 62 (2014) 72–92.[35] L. Wu, A. Jain, D.W. Brown, G.M. Stoica, S.R. Agnew, B. Clausen, et al.,

Acta Mater. 56 (2008) 688–695.[36] O. Muránsky, D.G. Carr, P. Šittner, E.C. Oliver, Int. J. Plasticity 25 (2009)

1107–1127.[37] L. Wu, S.R. Agnew, D.W. Brown, G.M. Stoica, B. Clausen, A. Jain, et al.,

Acta Mater. 56 (2008) 3699–3707.[38] D. Hull, D.J. Bacon, Introduction to Dislocations, Pergamon Press,

Oxford, 1984.[39] Y. Xiong, Y.Y. Jiang, Int. J. Fatigue 64 (2014) 74–83.

187C. Wang et al. / Journal of Magnesium and Alloys 4 (2016) 181–187